extern int module_id;
extern int modify_window;
extern char *partial_dir;
-extern struct exclude_list_struct server_exclude_list;
+extern struct filter_list_struct server_filter_list;
int sanitize_paths = 0;
{
int ret;
-#if HAVE_SOCKETPAIR
+#ifdef HAVE_SOCKETPAIR
ret = socketpair(AF_UNIX, SOCK_STREAM, 0, fd);
#else
ret = pipe(fd);
"ABCDEFGHIJKLMNOPQRSTUVWXYZ"
"0123456789"
",.-_=+@/") != strlen(*cmd)) {
- rprintf(FINFO, "\"%s\" ", *cmd);
+ rprintf(FINFO, "\"%s\" ", safe_fname(*cmd));
} else {
- rprintf(FINFO, "%s ", *cmd);
+ rprintf(FINFO, "%s ", safe_fname(*cmd));
}
}
rprintf(FINFO, "\n");
exit_cleanup(RERR_MALLOC);
}
-void overflow(char *str)
+void overflow_exit(char *str)
{
rprintf(FERROR, "ERROR: buffer overflow in %s\n", str);
exit_cleanup(RERR_MALLOC);
-int set_modtime(char *fname, time_t modtime)
+int set_modtime(char *fname, time_t modtime, mode_t mode)
{
- if (dry_run)
- return 0;
+#if !defined HAVE_LUTIMES || !defined HAVE_UTIMES
+ if (S_ISLNK(mode))
+ return 1;
+#endif
if (verbose > 2) {
rprintf(FINFO, "set modtime of %s to (%ld) %s",
- fname, (long)modtime,
+ safe_fname(fname), (long)modtime,
asctime(localtime(&modtime)));
}
+ if (dry_run)
+ return 0;
+
{
-#ifdef HAVE_UTIMBUF
+#ifdef HAVE_UTIMES
+ struct timeval t[2];
+ t[0].tv_sec = time(NULL);
+ t[0].tv_usec = 0;
+ t[1].tv_sec = modtime;
+ t[1].tv_usec = 0;
+# ifdef HAVE_LUTIMES
+ if (S_ISLNK(mode))
+ return lutimes(fname, t);
+# endif
+ return utimes(fname, t);
+#elif defined HAVE_UTIMBUF
struct utimbuf tbuf;
tbuf.actime = time(NULL);
tbuf.modtime = modtime;
return utime(fname,&tbuf);
-#elif defined(HAVE_UTIME)
+#elif defined HAVE_UTIME
time_t t[2];
t[0] = time(NULL);
t[1] = modtime;
return utime(fname,t);
#else
- struct timeval t[2];
- t[0].tv_sec = time(NULL);
- t[0].tv_usec = 0;
- t[1].tv_sec = modtime;
- t[1].tv_usec = 0;
- return utimes(fname,t);
+#error No file-time-modification routine found!
#endif
}
}
*
* Derived from GNU C's cccp.c.
*/
-static int full_write(int desc, char *ptr, size_t len)
+int full_write(int desc, char *ptr, size_t len)
{
int total_written;
/** Copy a file.
*
- * This is used in conjunction with the --temp-dir option */
+ * This is used in conjunction with the --temp-dir, --backup, and
+ * --copy-dest options. */
int copy_file(char *source, char *dest, mode_t mode)
{
int ifd;
if (verbose > 0) {
rprintf(FINFO,"renaming %s to %s because of text busy\n",
- fname, path);
+ safe_fname(fname), safe_fname(path));
}
/* maybe we should return rename()'s exit status? Nah. */
return fcntl(fd,F_SETLK,&lock) == 0;
}
-static int exclude_server_path(char *arg)
+static int filter_server_path(char *arg)
{
char *s;
- if (server_exclude_list.head) {
+ if (server_filter_list.head) {
for (s = arg; (s = strchr(s, '/')) != NULL; ) {
*s = '\0';
- if (check_exclude(&server_exclude_list, arg, 1) < 0) {
+ if (check_filter(&server_filter_list, arg, 1) < 0) {
/* We must leave arg truncated! */
return 1;
}
char **argv = *argv_ptr;
int argc = *argc_ptr;
int maxargs = *maxargs_ptr;
-#if !(defined(HAVE_GLOB) && defined(HAVE_GLOB_H))
+#if !defined HAVE_GLOB || !defined HAVE_GLOB_H
if (argc == maxargs) {
maxargs += MAX_ARGS;
if (!(argv = realloc_array(argv, char *, maxargs)))
if (!*s)
s = ".";
s = argv[argc++] = strdup(s);
- exclude_server_path(s);
+ filter_server_path(s);
#else
glob_t globbuf;
- int i;
if (maxargs <= argc)
return;
s = strdup(s);
memset(&globbuf, 0, sizeof globbuf);
- if (!exclude_server_path(s))
+ if (!filter_server_path(s))
glob(s, 0, NULL, &globbuf);
if (MAX((int)globbuf.gl_pathc, 1) > maxargs - argc) {
maxargs += globbuf.gl_pathc + MAX_ARGS;
if (globbuf.gl_pathc == 0)
argv[argc++] = s;
else {
- int j = globbuf.gl_pathc;
+ int i;
free(s);
- for (i = 0; i < j; i++) {
+ for (i = 0; i < (int)globbuf.gl_pathc; i++) {
if (!(argv[argc++] = strdup(globbuf.gl_pathv[i])))
out_of_memory("glob_expand_one");
}
return cnt;
}
-/* Turns multiple adjacent slashes into a single slash; gets rid of "./"
- * elements; collapses ".." elements except for those at the start of the
- * string; removes a trailing slash. If the resulting name would be empty,
- * change it into a ".". */
-unsigned int clean_fname(char *name)
+/* Turns multiple adjacent slashes into a single slash, gets rid of "./"
+ * elements (but not a trailing dot dir), removes a trailing slash, and
+ * optionally collapses ".." elements (except for those at the start of the
+ * string). If the resulting name would be empty, change it into a ".". */
+unsigned int clean_fname(char *name, BOOL collapse_dot_dot)
{
char *limit = name - 1, *t = name, *f = name;
int anchored;
continue;
}
/* collapse ".." dirs */
- if (f[1] == '.' && (f[2] == '/' || !f[2])) {
+ if (collapse_dot_dot
+ && f[1] == '.' && (f[2] == '/' || !f[2])) {
char *s = t - 1;
if (s == name && anchored) {
f += 2;
f += 2;
continue;
}
- *t++ = *f++;
- *t++ = *f++;
- limit = t;
+ limit = t + 2;
}
}
while (*f && (*t++ = *f++) != '/') {}
* The rootdir string contains a value to use in place of a leading slash.
* Specify NULL to get the default of lp_path(module_id).
*
- * If depth is > 0, it is a count of how many '..'s to allow at the start
- * of the path.
+ * If depth is >= 0, it is a count of how many '..'s to allow at the start
+ * of the path. Use -1 to allow unlimited depth.
*
- * We call clean_fname() to clean up the path, but we preserve a trailing
- * slash because that is sometimes significant on command-line arguments.
- */
+ * We also clean the path in a manner similar to clean_fname() but with a
+ * few differences:
+ *
+ * Turns multiple adjacent slashes into a single slash, gets rid of "." dir
+ * elements (INCLUDING a trailing dot dir), PRESERVES a trailing slash, and
+ * ALWAYS collapses ".." elements (except for those at the start of the
+ * string up to "depth" deep). If the resulting name would be empty,
+ * change it into a ".". */
char *sanitize_path(char *dest, const char *p, const char *rootdir, int depth)
{
char *start, *sanp;
}
if (*p == '.' && p[1] == '.' && (p[2] == '/' || p[2] == '\0')) {
/* ".." component followed by slash or end */
- if (depth > 0 && sanp == start) {
- /* allow depth levels of .. at the beginning */
- --depth;
- *sanp++ = *p++;
- *sanp++ = *p++;
- /* move virtual beginning to leave .. alone */
- start = sanp;
- continue;
- }
- p += 2;
- if (sanp != start) {
- /* back up sanp one level */
- --sanp; /* now pointing at slash */
- while (sanp > start && sanp[-1] != '/') {
- /* skip back up to slash */
- sanp--;
+ if (depth <= 0 || sanp != start) {
+ p += 2;
+ if (sanp != start) {
+ /* back up sanp one level */
+ --sanp; /* now pointing at slash */
+ while (sanp > start && sanp[-1] != '/') {
+ /* skip back up to slash */
+ sanp--;
+ }
}
+ continue;
}
- continue;
+ /* allow depth levels of .. at the beginning */
+ depth--;
+ /* move the virtual beginning to leave the .. alone */
+ start = sanp + 3;
}
/* copy one component through next slash */
while (*p && (*sanp++ = *p++) != '/') {}
curr_dir_len += len;
}
- curr_dir_len = clean_fname(curr_dir);
+ curr_dir_len = clean_fname(curr_dir, 1);
return 1;
}
return 1;
}
-/**
- * Return the filename, turning any newlines into '?'s. This ensures that
- * outputting it on a line of its own cannot generate an empty line. This
- * function can handle only 2 names at a time!
- **/
-const char *safe_fname(const char *fname)
+/* Return the filename, turning any non-printable characters into escaped
+ * characters (e.g. \n -> \012, \ -> \\). This ensures that outputting it
+ * cannot generate an empty line nor corrupt the screen. This function can
+ * return only MAX_SAFE_NAMES values at a time! The returned value can be
+ * longer than MAXPATHLEN (because we may be trying to output an error about
+ * a too-long filename)! */
+char *safe_fname(const char *fname)
{
- static char fbuf1[MAXPATHLEN], fbuf2[MAXPATHLEN];
- static char *fbuf = fbuf2;
- char *nl = strchr(fname, '\n');
-
- if (!nl)
- return fname;
-
- fbuf = fbuf == fbuf1 ? fbuf2 : fbuf1;
- strlcpy(fbuf, fname, MAXPATHLEN);
- nl = fbuf + (nl - (char *)fname);
- do {
- *nl = '?';
- } while ((nl = strchr(nl+1, '\n')) != NULL);
+#define MAX_SAFE_NAMES 4
+ static char fbuf[MAX_SAFE_NAMES][MAXPATHLEN*2];
+ static int ndx = 0;
+ int limit = sizeof fbuf / MAX_SAFE_NAMES - 1;
+ char *t;
+
+ ndx = (ndx + 1) % MAX_SAFE_NAMES;
+ for (t = fbuf[ndx]; *fname; fname++) {
+ if (*fname == '\\') {
+ if ((limit -= 2) < 0)
+ break;
+ *t++ = '\\';
+ *t++ = '\\';
+ } else if (!isprint(*(uchar*)fname)) {
+ if ((limit -= 4) < 0)
+ break;
+ sprintf(t, "\\%03o", *(uchar*)fname);
+ t += 4;
+ } else {
+ if (--limit < 0)
+ break;
+ *t++ = *fname;
+ }
+ }
+ *t = '\0';
- return fbuf;
+ return fbuf[ndx];
}
/**
p1 = p2 = "";
else {
p1 = curr_dir;
- p2 = "/";
+ for (p2 = p1; *p2 == '/'; p2++) {}
+ if (*p2)
+ p2 = "/";
}
if (module_id >= 0) {
m1 = " (in ";
m2 = lp_name(module_id);
m3 = ")";
- if (*p1) {
+ if (p1 == curr_dir) {
if (!lp_use_chroot(module_id)) {
char *p = lp_path(module_id);
if (*p != '/' || p[1])
p1 += strlen(p);
}
- if (!*p1)
- p2++;
- else
- p1++;
}
- else
- fn++;
} else
m1 = m2 = m3 = "";
fn = fname;
if ((int)pathjoin(t, sz, partial_dir, fn) >= sz)
return NULL;
- if (server_exclude_list.head
- && check_exclude(&server_exclude_list, partial_fname, 0) < 0)
- return NULL;
+ if (server_filter_list.head) {
+ static int len;
+ if (!len)
+ len = strlen(partial_dir);
+ t[len] = '\0';
+ if (check_filter(&server_filter_list, partial_fname, 1) < 0)
+ return NULL;
+ t[len] = '/';
+ if (check_filter(&server_filter_list, partial_fname, 0) < 0)
+ return NULL;
+ }
return partial_fname;
}
dir = partial_fname;
if (create) {
STRUCT_STAT st;
-#if SUPPORT_LINKS
int statret = do_lstat(dir, &st);
-#else
- int statret = do_stat(dir, &st);
-#endif
if (statret == 0 && !S_ISDIR(st.st_mode)) {
if (do_unlink(dir) < 0)
return 0;
return malloc(size * num);
return realloc(ptr, size * num);
}
+
+/* Take a filename and filename length and return the most significant
+ * filename suffix we can find. This ignores suffixes such as "~",
+ * ".bak", ".orig", ".~1~", etc. */
+const char *find_filename_suffix(const char *fn, int fn_len, int *len_ptr)
+{
+ const char *suf, *s;
+ BOOL had_tilde;
+ int s_len;
+
+ /* One or more dots at the start aren't a suffix. */
+ while (fn_len && *fn == '.') fn++, fn_len--;
+
+ /* Ignore the ~ in a "foo~" filename. */
+ if (fn_len > 1 && fn[fn_len-1] == '~')
+ fn_len--, had_tilde = True;
+ else
+ had_tilde = False;
+
+ /* Assume we don't find an suffix. */
+ suf = "";
+ *len_ptr = 0;
+
+ /* Find the last significant suffix. */
+ for (s = fn + fn_len; fn_len > 1; ) {
+ while (*--s != '.' && s != fn) {}
+ if (s == fn)
+ break;
+ s_len = fn_len - (s - fn);
+ fn_len = s - fn;
+ if (s_len == 4) {
+ if (strcmp(s+1, "bak") == 0
+ || strcmp(s+1, "old") == 0)
+ continue;
+ } else if (s_len == 5) {
+ if (strcmp(s+1, "orig") == 0)
+ continue;
+ } else if (s_len > 2 && had_tilde
+ && s[1] == '~' && isdigit(*(uchar*)(s+2)))
+ continue;
+ *len_ptr = s_len;
+ suf = s;
+ if (s_len == 1)
+ break;
+ /* Determine if the suffix is all digits. */
+ for (s++, s_len--; s_len > 0; s++, s_len--) {
+ if (!isdigit(*(uchar*)s))
+ return suf;
+ }
+ /* An all-digit suffix may not be that signficant. */
+ s = suf;
+ }
+
+ return suf;
+}
+
+/* This is an implementation of the Levenshtein distance algorithm. It
+ * was implemented to avoid needing a two-dimensional matrix (to save
+ * memory). It was also tweaked to try to factor in the ASCII distance
+ * between changed characters as a minor distance quantity. The normal
+ * Levenshtein units of distance (each signifying a single change between
+ * the two strings) are defined as a "UNIT". */
+
+#define UNIT (1 << 16)
+
+uint32 fuzzy_distance(const char *s1, int len1, const char *s2, int len2)
+{
+ uint32 a[MAXPATHLEN], diag, above, left, diag_inc, above_inc, left_inc;
+ int32 cost;
+ int i1, i2;
+
+ if (!len1 || !len2) {
+ if (!len1) {
+ s1 = s2;
+ len1 = len2;
+ }
+ for (i1 = 0, cost = 0; i1 < len1; i1++)
+ cost += s1[i1];
+ return (int32)len1 * UNIT + cost;
+ }
+
+ for (i2 = 0; i2 < len2; i2++)
+ a[i2] = (i2+1) * UNIT;
+
+ for (i1 = 0; i1 < len1; i1++) {
+ diag = i1 * UNIT;
+ above = (i1+1) * UNIT;
+ for (i2 = 0; i2 < len2; i2++) {
+ left = a[i2];
+ if ((cost = *((uchar*)s1+i1) - *((uchar*)s2+i2)) != 0) {
+ if (cost < 0)
+ cost = UNIT - cost;
+ else
+ cost = UNIT + cost;
+ }
+ diag_inc = diag + cost;
+ left_inc = left + UNIT + *((uchar*)s1+i1);
+ above_inc = above + UNIT + *((uchar*)s2+i2);
+ a[i2] = above = left < above
+ ? (left_inc < diag_inc ? left_inc : diag_inc)
+ : (above_inc < diag_inc ? above_inc : diag_inc);
+ diag = left;
+ }
+ }
+
+ return a[len2-1];
+}